On Tue, Aug 16, 2016 at 05:18:24PM +0100, Catalin Marinas wrote: > On Mon, Aug 15, 2016 at 10:45:09AM -0700, Kees Cook wrote: > > On Mon, Aug 15, 2016 at 3:47 AM, Catalin Marinas > > <catalin.marinas@xxxxxxx> wrote: > > > On Fri, Aug 12, 2016 at 11:23:03AM -0700, Kees Cook wrote: > > >> On Thu, Aug 11, 2016 at 10:44 AM, Catalin Marinas > > >> <catalin.marinas@xxxxxxx> wrote: > > >> > The ARMv8 architecture allows execute-only user permissions by clearing > > >> > the PTE_UXN and PTE_USER bits. However, the kernel running on a CPU > > >> > implementation without User Access Override (ARMv8.2 onwards) can still > > >> > access such page, so execute-only page permission does not protect > > >> > against read(2)/write(2) etc. accesses. Systems requiring such > > >> > protection must enable features like SECCOMP. > > >> > > >> So, UAO CPUs will bypass this protection in userspace if using > > >> read/write on a memory-mapped file? > > > > > > It's the other way around. CPUs prior to ARMv8.2 (when UAO was > > > introduced) or with the CONFIG_ARM64_UAO disabled can still access > > > user execute-only memory regions while running in kernel mode via the > > > copy_*_user, (get|put)_user etc. routines. So a way user can bypass this > > > protection is by using such address as argument to read/write file > > > operations. > > > > Ah, okay. So exec-only for _userspace_ will always work, but exec-only > > for _kernel_ will only work on ARMv8.2 with CONFIG_ARM64_UAO? > > Yes (mostly). With UAO, we changed the user access routines in the > kernel to use the LDTR/STTR instructions which always behave > unprivileged even when executed in kernel mode (unless the UAO bit is > set to override this restriction, needed for set_fs(KERNEL_DS)). > > Even with UAO, we still have two cases where the kernel cannot perform > unprivileged accesses (LDTR/STTR) since they don't have an exclusives > equivalent (LDXR/STXR). These are in-user futex atomic ops and the SWP > emulation for 32-bit binaries (armv8_deprecated.c). But these require > write permission, so they would always fault even when running in the > kernel. futex_atomic_cmpxchg_inatomic() is able to return the old value > without a write (if it differs from "oldval") but it doesn't look like > such value could leak to user space. If this was an issue, couldn't we add a dummy LDTR before the LDXR, and have the fixup handler return -EFAULT? Either way, this series looks technically fine to me: Reviewed-by: Will Deacon <will.deacon@xxxxxxx> but it would be good for some security-focussed person (Hi, Kees!) to comment on whether or not this is useful, given the caveats you've described. If it is, I can queue it for 4.9. Will -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx"> email@xxxxxxxxx </a>